TRANSPORTATION RESEARCH RECORD 1483 11

Effects of Federal Levees and Reservoirs on 1993 Stages in St. Louis

GARY R. DYHOUSE

A physical model of the Middle was used to perform· (3000 acres) of intense commercial and industrial development in a quantitative analyses of the effects of federal levees on flood heights. narrow strip along the shore. These units are shown on Contrary to some opinions by the general public, these tests showed that Figure 2. The seven agricultural levee units downstream from the today's stage-frequency relationship at the St. Louis gage is not greatly urban area are designed for a 2 percent annual chance (50-year different than that existing before the onset of changes to the river by humans, beginning in the 1820s. Federal levees result in increased flood recu~ence interval) event. These seven units provide partial heights upstream of the protected area, but these increases have been protection to more than 80,900 hectares (200,000 acres) of highly fully offset by reductions resulting from federal flood control res_ervoirs productive farmland. As seen, most of the Middle Mississippi River for all in recent years. The occurrence of the Great Flood of 1993 flood plain has a high level of protection. fuhher confirmed these findings and continued to demonstrate the value Throughout the 1970s and early 1980s every flood along the of the federal levees and reservoirs. Federal levees are estimated to have Middle Mississippi River resulted in complaints by local citizen and increased flood heights at St. Louis by 0.9 to 1.2 m (3 to 4 ft) during the. crest of the 1993 flood; however, federal reservoirs reduced flood levels environmental groups that the Corps had actually made flooding at St. Louis by an equal amount, counterbalancing the levee increases. worse by levee construction in the area (J,2). Extensive complaints An estimated 19 billion dollars in flood damage was preventep by the by residents of St. Charles County, located just upstream of the end federal levees and reservoirs throughout the Upper Mississippi and of levee system and between the junction of the Mississippi and Basins in 1993, a savings of more than the total Missouri rivers, resulted in the initiation and completion of qualita­ construction cost of the projects. The model tests ~lso found that the tive studies. These studies are described elsewhere (3,4). These key historical flood discharges of the I &44 ·and 1903 floods, used in qualitative studies demonstrated that federal levees had much less the design of the current levee system, may be overestimated by up to 33 percent. an effect than had previously been thought. However, additional more quantitative tests were needed to evaluate several scenarios and to better determine the effects of only the federal levee raises St. Louis, Missouri, is one of the best known river towns in the on flood heights near St. Louis. These tests proved invaluable as the world, popularized in Mark Twain's. books about life on the Mis­ Great Flood of 1993 unfolded, and the effect of levees on flood sissippi River. Although most of the city sits on high ground, well levels was a question asked by the news media from throughout the above maximum flood levels, low areas are occasionally threatened world. This paper describes these model tests and the results and by three large rivers that join a short dist~nce upstream of the city uses these tests to focus on the effects of federal flood reduction of St. Louis. The , the Upper Mississippi, and the Missouri projects on the Great Flood of 1993. Rivers form the "Father of Waters," the great Mississippi, which flows south, gathering the inflow from the Ohio, the Arkansas, the Red, and many other lesser rivers, on its long journey to the Gulf of PHYSICAL MODEL TESTS . Mexico. The river reach from the Missouri to the Ohio is designated as the Middle Mississippi River. It extends from the mouth of the In 1985-86, a series of physical model tests was performed for the Ohio (Km 0) to the mouth of the Missouri River (Km 313.8, or Mile St. Louis District by the Corps' Waterways Experiment Station 195), with downtown St. Louis located at Km 289.7 (Mile 180). To using the Mississippi Basin Model (MBM) in Clinton, Mississippi. reduce the frequent and repetitive flood damage in this reach, the St. It was proposed to isolate the effects of the federally raised levees Louis District of the Corps of Engineers had overseen construction on flooding by comparing flood elevations for today's conditions of levee raises in the 1950s and 1960s to the privately constructed with those in existence in 1820, before any effects by humans (5), Alton to Gale (Illinois) levees. The federal levee system through in 1940 (before any flood reduction work in the reach by the Corps this reach is shown in Figure 1. This system provides protection to of Engineers) and by comparing today's conditions with only the the urban and agricultural flood plain areas from Km 74 (Mile 46) private levees of 1940 (no federal levees). In addition, the MBM at Gale (just downstream of Cape Girardeau, Missouri) to Km 326.7 was to be used to estimate the historical peak discharges for the (Mile 203) at Alton. Four major urban levee units in the St. Louis. 1844 and 1903 floods. These two events, neither of which was area give protection from a river stage of 15.8 m (52 ft) on the gaged at St. Louis, play a large part in flood evaluations and levee St. Louis gage, a stage estimated as rarer than a 0.2 percent annual design on the Middle Mississippi. chan~e (500-year recurrence interval) event. Three of these units protect mo~e than 40,500 hectares (100,000 acres) in the urbanized Illin~~~ floodplain, and the fourth protects less than 1200 hec~ares MODEL ADJUSTMENT AND CALIBRATION

Gary R. Dyhouse, Hydrologic Engineering Section', Corps of Engineering These tests required extensive updating and modifications of the I 222 Spruce Street, St. Louis, Mo. 63103c2833. MBM. Today's conditions landward of the levees were fully incor- 12 TRANSPORTATION RESEARCH RECORD 1483

Illinois River

Cuivre River

ILLINOIS

MISSOURI

LEGEND:

Levee O Gage ALTON TO GALE . -r7r .Bluff LEVEE SYSTEM

FIGURE 1 Levee location: Alton to Gale, Illinois.

porated by adding the extensive railroad and highway fills across the of 1973, 1982, and 1983. After calibration was achieved, a series of floodplain, adding new bridges to the model and including the discharges (Table 1), ranging from a 10-year recurrence interval extensive flood plain urbanization in the St. Louis area. The model through the estimated 1844 peak discharge (rarer than a 500-year was changed to simulate 1940 conditions by installing only the pri­ recurrence interval event) was run for each of the four scenarios. vate levees in place at that time, as well as the vegetation patterns The results were surprising. and land use of 1940, based on period maps. Finally, the 1820 con­ dition was modeled by removing all levees and flood plain obstruc­ tions and adding the dense vegetation then prevalent throughout the GENERAL FEDERAL LEVEE EFFECTS flood plain. An excellent chart of the river valley, prepared in 1797 and showing the vegetation in the valley, was available to use as a To estimate the effects of levees, local interest groups in the area guide. The channel geometry was not adjusted because earlier stud­ had used gage records that showed higher recent levels of flooding ies (6) had found that the Middle Mississippi River showed similar compared with older records for the same discharge, ignoring the channel widths and surface areas as it does today. However, one effects of any other changes in the system between flood years. reach of the main channel was modified for 1820 conditions by Besides the effects of humans, the dynamics of the Mississippi re-installing a 13-km (8-mi) reach of the river, about 97 km (60 mi) River can result in significantly different river stages for a given dis­ downstream from St. Louis that had been cutoff naturally by the charge. Water temperature, river planforms, vegetation, and other river in 1888. Today's conditions (with federal levees) were exten­ features have caused stage changes of several feet for specific flood sively calibrated to known stage and discharge data for the floods discharges. Comparing gage records alone cannot address the Dyhouse 13

ILLINOIS

MISSOURI

LEGEND:

Major Highway Railroad Corp Levee Bluff Line

FIGURE 2 Levee location: St. Louis Area.

changes of river stage caused by only one variable: levees. The only depending on the discharge and location of flooding. Previous way of determining the effects of levees alone is through the use of rough analyses by private citizens, based on gage record compar­ a physical or analytical model in which all variables are held con­ isons only, had sought to show that these increases were 2.4 to 4 m stant except levees . .This was the objective in the series of MBM (8 to 13 ft). Furthermore, the increases caused by federal levee tests. The MBM tests showed that the federal levees (as expected) raises had been minimal for floods occurring before 1993, generally cause increases in flooding; however, these increases were from less a 0.3-m (I-ft) increase maximum. The simulated 1940 land use con­ than 0.1 m (4 in.) to 1.2 m (4 ft) compared with 1820 conditions, ditions showed that stage-discharge relationships had decreased 14 TRANSPORTATION RESEARCH RECORD 1483

.TABLE 1 Test Discharges and Frequencies at St. Louis

Discharge (ems) Today's Frequency Pre-Reservoir Frequency Average return interval (years)

19,540 10 5 22,220 20-25 10 26,200 50 20 28,890 100 40 34,550 300-400 100 36,820 500-1000 200

36,820 ems is the levee design discharge at St. Louis 1 m 3/sec = 35.71ft3/sec

compared with 1820 conditions: This change was evidently because GENERAL FEDERAL RESERVOIR EFFECTS of the extensive flood plain clearing for agriculture use that occurred continuously throughout the 1800s and early 1900s. An average Differences resulting from various land use and levee scenarios for reduction of about 0.7 m (2.3 ft) for moderate to severe flood dis­ stage-discharge relationships also do not include the effects of charges was shown by the model results. Flood elevation differ­ upstream reservoirs on the Upper Mississippi or Missouri River ences between today's with-levee river compared with the 1820 watersheds. The changes in discharge-frequency relationships at conditions showed increases of only 0.1 to 1.2 m (0.3 to 4 ft) in key gages throughout the Missouri and Mississippi River systems St. Louis. Although today's flood flows are constrained to a leveed were first developed in the 1950s through extensive evaluation and floodway, ranging from about 600 to 1000 m (2000 to 3300 ft) in routing of actual and hypothetical floods with and without several width for a several kilometer reach in St. Louis, the effects of this different reservoir scenarios. The adopted discharge-frequency conveyance loss are not as significant as previously thought. The relationships in St. Louis and at other nearby gages have been dense flood plain forests that were common in the 1800s evidently reevaluated on three occasions in the past 20 years and are consis­ also limited effective flow area outside the main channel. Upstream tent with statistical analyses performed on the data stream for and downstream of the narrow St. Louis reach, the average width of St. Lo~is. The differences between pre- and post-reservoir condi­ the floodway widens out to 2 to 3 times the ~idth in St. Louis. Fig­ tions are pronounced, generally decreasing peak discharges at ures 3 and 4 illustrates stage-discharge relationships at the St. Louis St. Louis by 15 percent or more. All significant floods on the Middle and Alton, Illinois, gages, respectively. The frequencies shown in Mississippi since the 1973 event have been reduced by an average Figures 3 and 4 are for existing conditions only and do not apply to of 1.5 m (5 ft) through upstream reservoir holdouts, as shown on the 1940 or 1820 time periods, before the construction of numerous Table 2. The largest reduction occurred during the flood of April flood control reservoirs that significantly reduced the discharges 1994. Peak stage at St. Louis for this event was estimated to have associated with a specific frequency. been up to 3 m (10 ft) higher without the reduction effects of fed-

ST. LOU IS GAGE ALTON GAGE KM 289.7 KM 326.3 136 ~---.------.-----,------,------, -e- EXISTING --- 1820 132 -9- 1940 -B- EXISTING W/O FED. LEVEES ...--..131 E '-"' z 130 z 0 0133 t-----i-___.__.___,.=------j::: ~ 129 r-----+----.,r­ <( > > w w1321-----+--+--N<+-,1U---~ _J ci 128 r-----1-7' w

127 1-----1------+

126 .____ _.______._ ___.._____ --'------' 130 ~--~--~--~ 15 20 25 30 35 40 15 20 25 30 35 40 DISCHARGE IN 1000 m3/sec DISCHARGE IN 1000 m 3/sec

FIGURE 3 Stage discharge relationship at St. Louis Gage. FIGURE 4 Stage discharge relationship at Alton, Illinois, Gage. Dyhouse 15

TABLE 2 Estimated Resevoir Effects at St. Louis

Approximate Decrease in Actual Approximate Stage Discharge Stage Stage With No Reservoirs Flood (m3/secl (m) (m) (m)

1979 3, 115 0.9-1.2 11.5 12.7 1982 2,550 0.6-0.9 12.0 12.8 1983 3,965 1.2-1.5 11. 1 12.3 1986 5,665 1.8-2.1 11.9 13.9 1993 9,005 0.9-1.2 15.1 16.2 1994 8,495* 2.4-3.0 11.2 13.9

Flood Stage (Bankfull) = 9.1 m (30 ft) 1 m = 3.28 ft, 1 ft3/sec = 0.0283 m 3/sec

*Stage reductions are much greater for this flood because the 1994 flood would still be contained by the levee system; the 1993 flood would have overtopped all Federal urban levees in the St. Louis area, limiting the stage reductions by reservoirs.

eral reservoirs. Because of the extensive system of reservoirs, mete­ • The frequency of the peak discharge is estimated to be a orological and climatological events occurring today will result in 150- to 200-year exceedance frequency at St. Louis. Some upstream less water reaching the Missouri and Mississippi rivers at the crest stations have estimated the flood· as rarer than the 500-year of a flood compared with conditions before reservoir construction, exceedance frequency event. which commenced in earnest in about 1940. When one includes both levee and reservoir effects, today's stage-frequency relation­ Despite these staggering statistics, the performance of federally ships are not much different than those of the early 1820s. constructed levees, flood walls, and flood control reservoirs was out­ standing and prevented most potential damage. For instance, flood GREAT FLOOD OF 1993 damage in the St. Louis District alone is estimated as $1.4 billion; damages prevented by the. federal flood reduction components are The 1993 flood on the Mississippi and Missouri rivers in the Upper estimated at $5.4 billion. Thus, about an 80 percent reduction in Midwest broke all records for stage, volume, peak discharge, dura­ potential flood damages along the 483 km (300 mi) of the Missis­ tion, and frequency for more than 1610 km (1000 mi) of the two sippi River and tributaries in the St. Louis district was achieved. rivers and for many of their major tributaries. A sense of the mag­ nitude of the Great Flood of 1993 can be obtained by reviewing some of the statistics at the St. Louis gage in downtown St. Louis. Performance of Levees

• The peak stage exceeded the previous flood of record by nearly The flood overtopped or broke 1082 of 1571 levees in the watersheds 2 m (6.2 ft). of the Mississippi and Missouri rivers, upstream from the mouth of • The discharge of30,280 m3/sec (1,070,000 ft 3/sec) is the great­ the . Most of these levees were constructed by private or est discharge ever measured during more than 130 years of site data, nonfederal entities and protected primarily agricultural areas, with exceeding the previous high by 26 percent. protection levels from a 20 (5-year) to 5 percent (20-year) annual • Flood elevations exceeded the flood stage (bankfull capacity) of chance event. Given the severity of the Great Flood of 1993, it 9.15 m (30 ft) for 80 consecutive days during the main portion of the should not be surprising that 1043 of the 1345 private and nonfed­ flood and for 148 days during the calendar year. The previous record eral levees could not prevent the river from flooding the agricultural was 77 days above flood stage, both consecutively and annually. areas. Of the 226 federal levees, only 39 succumbed to the river. • .Durations of flooding at hig~ stages were unprecedented. The These 39 federal levees also were designed to protect primarily agri­ flood was 3 m (10 ft) or more over flood stage for 36 days, exceeded cultural areas and offered protection ranging from a 5 percent annual the "50-year flood" stage for 23 days, and exceeded the "100-year chance (20-year) to a 2 percent annual chance (50-year) event. Only flood" stage for 8 days. Before 1993, there were only 12 days total one of these levees failed, with the definition of a failure meaning the in the entire period of record, dating back to 1861, that exceeded levee broke before the river exceeded the design river stage for the flood stage by 3 m (10 ft) or more. levee. Nearly all flooding of protected lands behind federal levees • The total water volume passing St. Louis during the main body occurred through overtopping of the levee; only four units filled of the flood, from June 26 to September 13, was about 1.382 X 10 11 before overtopping. Of these four units, two broke because ofunder­ m3 (112 million acre-ft). This volume could fill a box 1.61 X 1.61 seepage, piping or scour when the river stage was greater than the X 53.3 km high (1 X 1 X 33.1 mi high). design river stage but was less than the levee crest elevation. One • The average daily stage for the 1993 calendar year approxi­ filled because of human failure to operate a closure structure at a rail­ mates the average annual high stage, on the basis of more than road opening, and one levee actually failed, broke before floodwaters 130 years of data. reached as high as the design stage. All levees and floodwalls 16 TRANSPORTATION RESEARCH RECORD 1483 designed for floods of the magnitude of the 1993 event performed as August 1 when the first of two large downstream agricultural levees expected, and no flooding resulted to protected areas. In addition, was overtopped and filled. These two levees were both designed to several levees designed for events less severe than the 1993 flood provide protection from the 2 percent annual chance (50-year recur­ held through extraordinary floodfighting measures. Overall, the per­ rence interval) event and did not actually overtop until river stages formance of federal levees and flood walls was more than one could had reached about 1.2 m (4 ft) higher than the levee design river reasonably expect, given the severity of the Great Flood of 1993. stage. Had no overtopping occurred, it is estimated the St. Louis These findings are in agreement with those of the Report of the Inter­ stage would have crested about 0.3 m (1 ft) higher. With no levees agency Floodplain Management Review Commitee (I), which existing on the Middle Mississippi River, it is estimated that river found that levees greatly reduced the actual flood damages but had stages at St. Louis would have been 0.9 to 1.2 m (3 to 4 ft) lower. only minor overall effects on flood stages throughout the Mississippi These figures were independently confirmed by a university study and Missouri River systems. commissioned by a local newspaper and are similar to other find­ ings (I 1). Although a further reduction of 1 to 1.2 m appears large, one has to evaluate the significance of this reduction. If all levees Performance of Reservoirs were removed, the river would still have crested at least 4.9 m (16 ft) over flood stage. However, then the river would have flowed The Federal Government operates more than 70 flood control reser­ from bluff to bluff through the St. Louis area, reaching up to 16 km voirs in the Upper Mississippi and Missouri watersheds, with most (10 mi) wide at some locations. Several billion dollars in additional of these ·structures in the Missouri River Basin. These dams and . damages would have been caused by the flood. In addition, there reservoirs range from huge structures on the Missouri River main would not be a significant reduction in damages elsewhere, because stem in the Dakotas and Montana to small structures well up in the of the limited flood reductions attributable to levee removal. As headwaters of tributary streams. The rainfall of 1993 was very wide­ mentioned previously, federal flood control reservoirs reduced spread, so most of these structures retained significant volumes of flooding in St. Louis by about 0. 9 to 1.2 m (3 to 4 ft). Thus, the flood water, resulting in important crest reductions on both major rivers. reduction effects of reservoirs again offset the local effects of levees Current estimates of reservoir effects on the Great Flood of 1993 in St. Louis. This offsetting effect has been noted for every signifi­ show about a 0.9 to 1.2 m (3 to 4 ft) reduction in St. Louis. This lower cant flood in St. Louis since 1973. Table 2 illustrates the effects of stage represents a peak discharge reduction of about 9000 m3/sec reservoirs on St. Louis flood peaks in recent years. (over 320,000 ft3/sec); the peak discharge at St. Louis would have been over 39,000 rri.3/sec (1,391,000 ft3/sec) instead of the actual 30,280 m3/sec (1,070,000 ft3/sec ). All four urban levees in the HISTORICAL DISCHARGE EVALUATION St. Louis area probably would have been overtopped during the 1993 flood without the upstream reservoir control of this event. Through­ The MBM was also used to determine a peak discharge that would out the Middle Mississippi, reservoir reductions of crest stages reconstitute historical highwater marks from the 1844 and 1903 ranged from 0.9 to 1.6 m (3 to 5.1 ft). Significant reservoir reduc­ floods, under the appropriate land use conditions. These two floods tions occurred throughout the Missouri and Mississippi rivers, also have estimated peak discharges of 36,~20 m3/sec ( 1,300,000 ft 3/sec) an important finding (7) that indicated that federal levees and reser­ and 28,860 m3/sec (1,019,000 ft 3/sec), respectively, in St. Louis. voirs prevented more than $19 billion in flood damage during the The 1844 discharge is the basis for urban levee design heights in the 1993 event. As indicated by the Corps of Engineers (8), reservoir St. Louis reach. Neither flood discharge was measured in St. Louis. reductions along the Missouri River are estimated as 1.8 m (6 ft) in After the 1903 flood, an estimate of this flood was made for Sioux City, ; 1.5 to 2.4 m (5 to 8 ft) in Omaha, ; and St. Louis, based on measurements at gages 112 and 215 km (70 and 0.9 m (3 ft) in City, Missouri, and points downstream. Upper 134 mi) downstream conducted with the crude instruments avail­ Mississippi River reservoirs are few, and the reductions are less able at the time. Before 1930, velocity measurements were often upstream of the mouth of the Missouri River, generally 0.1 to 1.2 m made using floats (6). Float measurements have the tendency to (0.3 to 4 ft) along the Missouri and Illinois border. overestimate river velocity and thus the total discharge. The over­ Contrary to some opinions, federal flood control projects have estimate becomes progressively worse as flood discharge increases. proved their value conclusively. During this century, total federal The 1903 peak discharge value at St. Louis was estimated by expenditures for structural flood control features have been increasing (instead of decreasing) the discharge measured 113 km estimated at $25 billion to $30 billion (9). Since 1983, Corps of (70 mi) downstream with floats by about 13 percent, which appears Engineers projects alone have prevented $170 billion in damages unreasonable. The 1844 discharge at St. Louis was then estimated (10). A strong case can be made for the value of flood reduction through ratioing the downstream gage results for the 1903 flood projects, showing that the entire system pays for itself about every upward. Not included in this estimate was the fact that the river had 18 months, based on the last 10 years of data. shortened itself between th~ two floods through a natural cutoff in 1888, affecting both the slope and the channel storage available as well as the significant changes in flood plain land use between the Joint Effects at St. Louis two periods. There has long been a serious question about the valid­ ity of the St. Louis peak discharge estimates of 1844 and 1903 For the 1993 flood peak discharge of 30,280 m3/sec (1,070,000 because of the significantly changed river conditions between 1844 ft3/sec), the model tests suggested that the levees would result in a and 1903 and the likely overestimate of flood discharges. The MBM maximum stage increase of 1.2 to 1.5 m (4 to 5 ft), comparing 1940 tests found that flows much less than these official estimates gave a conditions with today's. The difference is about 0.3 m (1 ft) less if reasonable match of 1844 highwater marks under the 1820 land use comparing 1820 conditions, before the start of humans' modifica­ condition. Model flows of about 24,640 m3/sec (870,000 ft3/sec) and tions. The peak stage occurred in St. Louis from 0900-1000 hr on 22,090 m3/sec (780,000 ft 3/sec) for the 1844 and 1903 floods, Dy house 17 respectively, in St. Louis resulted in the matching of 12 to 15 known SUMMARY highwater marks (within 1 ft or less) available for both historic floods. These model peak discharge values are about 33 and 23 per­ Although precise knowledge of the actual channel and floodplain cent less, respectively, than current official estimates of the 1844 conditions in 1820 will never be known, the series of tests per­ and 1903 flood peaks at St. Louis and are far below the measured formed for conditions as they were believed to exist in 1820, in peak discharge of August 1, 1993 (30,280 m3/sec or 1,070,000 1940, and for today have shown far less negative effects of federal ft3/sec). Both Corps of Engineers and United States Geological Sur­ levees than had previously been estimated. The federal flood reduc­ vey personnel are further reviewing these results and the historical tion program for the Mississippi River, which has been ongoing for estimates for St. Louis. This review may eventually lead to a revi­ over 50 years, has accomplished a great deal for areas protected by sion of the historical records. The lower flows found by using the levees. Federal reservoirs, an important part of the flood reduction MBM are similar to floods occurring in the 1970s and 1980s. The program, have offset the negative results on unprotected areas in the April 1973 flood (24;130 m3/sec or 852,000 ft 3/sec) held the record St. Louis area caused by federal levees. The stage-frequency rela­ for peak stages throughout much of the Middle Mississippi before tionships along the Middle Mississippi River are little different 1993. Peak discharge of the 1973 event may have been within about under today's conditions compared with the estimated conditions of 2 percent of the actual 1844 peak. It seems apparent that the peak 1820. The occurrence of the Great Flood of 1993 served to verify discharge of the 1993 flood is the largest fl.ow rate seen in the these earlier studies on federal levee effects near St. Louis and to Middle Mississippi River Valley since before colonial settlers demonstrate the value of the federal flood protection program. observed the first significant flood in 1785. An earlier version of this paper, with the same title, was written in December 1993 and was presented at the American Institute of Hydrology Convention' in April 1994. This paper updates many of CONCLUSIONS the statistics cited in the earlier version and includes additional and more accurate data from the 1993 flood than was available in The main conclusions resulting from the series of tests and from the December 1993. Therefore, this paper supersedes the earlier version. Great Flood of 1993 include

• Stage-frequency relationships for today's conditions are not REFERENCES greatly different from those for the 1820 time frame, when humans first began clearing the bankline and floodplain. I. Belt, C. B., Jr. The 1973 Flood and Man's Constriction of the Missis­ • The clearing and draining of the Middle Mississippi floodplain sippi River. Science, Vol. 189, Aug. 1975, pp. 681-684. during the 1800s and early 1900s caused a significant decrease in 2. Stevens, M. A., S. A. Schumm, and D. A. Simons. Man Induced stage for a given flood discharge. Changes of Middle Mississippi River. ASCE Hydraulics Division • Federal levees alone result in a 0.1 to 1.2 m (0.3 to 4 ft) Journal, Proceedings Paper 11281, 1975. increase in flood heights for a given discharge in St. Louis compared 3. Dyhouse, G. R. Comparing Flood-Stage Discharge Data-Be Careful! Hydrology and Hydraulics in the Small Computer Age. Proc., with 1820 conditions, less than one-half the amount estimated by Hydraulics Division Specialty Conference, Orlando, Fla., Aug. 1985, private groups and individuals in the past. pp. 73-78. • Levees protecting the urban areas near St. Louis during the 4. Dyhouse, G. R. Levees at St. Louis: More Harm Than Good? Hydrology 1993 flood caused an increase of about 0.9 to 1.2 m (3 to 4 ft) in and Hydraulics in the Small Computer Age. Proc., Hydraulics Division Specialty Conference, Orlando, Fla., Aug. 1985, pp. 390-395. flood levels for several kilometers upstream. However, federal flood 5. Shreve, H. M. Annual Report of Work Done in the Improvement of the control reservoirs reduced flood levels at St. Louis by about the Navigation of the Ohio and Mississippi Rivers, through 30 September same amount, offsetting the adverse effects of levees on unpro­ 1830. U;S. War Department, Oct. 1830. tected areas. 6. Westphal, F. A., and S. P. Clemence. SLD Potamology Study (S-7). • Upstream flood control reservoirs fully offset the effects of fed­ Institute of River Studies, University of Missouri-Rolla, Dec. 1976. 7. Sharing the Challenge: Floodplain Management Into the 21st Century. eral levees for floods on the Middle Mississippi River in St. Louis. Report of lnteragency Floodplain Management Review Committee to • Structural flood control systems of levees, fl.oodwalls, and Administration Floodplain Management Task Force, Washington, reservoirs greatly reduced actual flood damages during the 1993 D.C., June 1994. flood. An estimated $19 billion in damages was prevented by fed­ 8. The Great Flood of 1993 Post-Flood Report. Corps of Engineers, North Central Division. Main Report, Sept. 1994. eral flood protection projects. 9. Williams, P. B. Flood Control vs. Flood Management. Civil Engineer • If the meteorological and climatological events of 1993 had Magazine, May 1994. occurred in the early 1800s, the crest stages in St. Louis likely 10. Annual Flood Damage Report to Congress for Fiscal Year 1993. would have been similar to the actual stages of 1993. U.S. Army Corps of Engineers, April 1994. • Published flood discharges along the Middle Mississippi 11. Science for Floodplain Management Into the 21st Centu·ry, Part V. Report of Interagency Floodplain Management Review Committee to before about 1930, after which time fl.oat measurements were no Administration Floodplain Management Task Force, Washington, longer used, appear to be seriously overestimated. The key floods D.C., June 1994. of 1844 and 1903 appear to be 33 and 23 percent overestimated. The 1973 flood may be the third highest discharge on record, after the This paper represents the findings and opinions of the author and not 1844 event and the record 1993 flood. necessarily those of the Corps of Engineers.